Test of Neural Network Techniques using Simulated Dual-Band Data of LEO Satellites

Abstract

Dual-band, multi-pass simulations of low Earth orbit (LEO) satellites are used to train a feedforward neural network to recognize different classes of resident space objects (RSO). Simulated data allow for a controllable and diverse set of inputs necessary to test our methods, especially at the initial phase of the evaluation, while avoiding the problems and expense associated with real data collection from ground-based facilities. Simulation software is used to generate signatures in two visible bands for satellites exhibiting typical bus-types, materials combinations and methods of stabilization. Orbits and observational parameters are generated from the relevant statistical distribution of the orbital parameters obtained from the Space Surveillance Network, and stabilization is simulated external to the framework of the software used to calculate signatures. We examine various pre-processing schemes that combine temporal, spectral and solar phase angle (SPA) information from non-glinting signatures into vectors that can be used as inputs for our classifier. A metric that assigns a proxy signal-to-noise ratio to each neural network output is introduced to determine the confidence level of each result.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 2010
Accession Number
ADA531683

Entities

People

  • Anthony V. Dentamaro
  • Kimberly R. Knobel
  • Phan D. Dao

Organizations

  • Air Force Research Laboratory

Tags

DTIC Thesaurus Topics

  • Artificial Satellites
  • Computer Programs
  • Earth Orbits
  • Ground Based
  • Information Science
  • Low Earth Orbits
  • Materials
  • Neural Networks
  • Orbits
  • Pattern Recognition
  • Radiant Intensity
  • Resident Space Objects
  • Simulations
  • Space Objects
  • Space Surveillance
  • Spacecraft
  • Spacecraft Orbits

Readers

  • Computational Linguistics
  • Phased Array Antenna Design.
  • Space Exploration and Orbital Mechanics.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Neural Networks
  • Space